Chapter 3: Diegetic UI & AR Overlays; AI/Autonomy Modes
Created by Sarah Choi (prompt writer using ChatGPT)
Diegetic UI & AR Overlays; AI/Autonomy Modes — Cockpits, Bridges & UX (for Vehicle Concept Artists)
Why Diegetic UI + AR Change the Cabin
Diegetic UI uses the vehicle’s own architecture—pillars, bezels, seams, glass—as the interface. Augmented reality (AR) overlays information onto the outside world at an apparent distance. Together they move the UX from screens to space. For concept‑side artists, these tools let you paint intent directly on structure and sightlines; for production‑side artists, they impose hard constraints on optics, lighting, latency, redundancy, and fail‑safe behavior. This article links human factors, HUD fundamentals, diegetic elements, AR registration, and autonomy mode design so your cabins feel inevitable and buildable.
Human‑Factor Ground Rules
Start with people: define the eyebox volume, seat ranges, and postures. Place primary cues within the operator’s forward visibility cone and set their apparent focal distance near the outside world (2–6 m for cars; optical infinity for aircraft/spacecraft). Place secondary cues on structure (A‑pillars, yoke rim, beltline) at near‑world focal distances (0.6–1.0 m) so refocus time stays short. Respect night vision and glove use: brightness floors, color use, and tactile affordances still matter when AR is down or eyes are dark‑adapted.
Diegetic UI: Interface as Architecture
Diegetic components include edge‑lit trim, pillar ladders, beltline light bars, rim LEDs, illuminated seams, and projected glyphs on surfaces. Use them to echo state (gears, flaps, drive modes), proximity (parking, docking), or checklists (latch closed, lock armed). Good diegetic rules: avoid specular glare; keep luminance contrast readable in sun; group functions by location (all drive‑mode signifiers at the yoke, safety at pillars). Provide haptic anchors (knurls, ridges) so hands find controls without sight. Production notes: IP ratings, thermal paths for LEDs/laser diodes, dimming curves, replaceable light engines, and service access that doesn’t sever structural rings.
AR Overlays: Optics, Registration, and Occlusion
AR must land in the right place at the right depth. That means:
- Optics & Focal Distance: Combiner HUDs or waveguides project at fixed or variable depth. If fixed, choose mid‑field so refocus cost is low; if variable (multi‑plane/varifocal), restrict to critical cues.
- Registration: Align graphics to the world with a calibrated sensor stack (IMU, cameras/lidar, wheel/air data). Use world‑locked elements for conformal cues (lane edges, runway aiming point) and screen‑locked for widgets.
- Occlusion: Never hide real hazards. Use depth masks so virtual objects dip behind real ones; use priority fading when line‑of‑sight is busy; indicate confidence as translucency or error bars.
- Latency & Jitter: Motion‑to‑photon delay must be below human thresholds (ideally <20 ms for head‑coupled cues). Stabilize with predictive filters and limit high‑frequency motion of graphics.
Information Hierarchy: What Lives Where
Keep a three‑tier model: world layer (conformal path, hazards, navigation), glass layer (speed/tape, pitch/roll, warnings), and structure layer (diegetic status bands, mode authority). Under workload, declutter automatically; under failure, fallback to minimal symbology. For bridges, put docking cues at window corners and range ladders up the pillars; for ground vehicles, project apex and braking arcs onto the road; for VTOL, show velocity vector and landing disc with clearance ring.
Color, Brightness, and Contrast
Adopt a semantic palette (warnings, cautions, advisories, guidance) and keep it consistent with physical annunciators. Match photopic/scotopic ranges with large dimming ratio (10,000:1) and auto‑adapt to ambient. Avoid pure red in NVG contexts; prefer cyan/white for conformal lines against night scenes. Provide anti‑washout strategies: boosted edge contrast, drop‑shadows, and polarization‑aware coatings.
Interaction Models: Hands, Voice, Gaze, and Gestures
AR invites touch‑in‑air, gaze dwell, and voice. In vehicles, glove‑safe and vibration‑tolerant inputs win. Combine gaze + confirm (look then click), voice + diegetic echo (spoken command mirrored as a pillar tag), and physical anchors for critical functions. Always provide hard controls for safety‑critical actions; AR augments, it does not replace.
Autonomy & Authority: Modes That Don’t Confuse
Define authority states clearly: Manual, Assist, Supervised Autonomy, Full Autonomy. Represent mode with a persistent diegetic signifier (thin band around yoke rim or beltline) plus a HUD badge. On transition, provide multimodal cues: haptic notch at the control, short tone, HUD banner. Show who’s driving at all times and where the system’s attention is (detected path, tracked hazards). Offer handover countdowns with obvious affordances (grasp wheel, press accept). In degraded states, switch to limp‑home UI: bigger fonts, minimal overlays, and explicit limits.
Explanations & Trust: Make the AI Legible
Expose intent (“planning left pass”), capability limits (“low map confidence”), and reason for alerts (“occluded pedestrian”). Use confidence meters or halo thickness on AR elements to telegraph certainty. Provide a timeline scrub after incidents (what the system saw, chose, and did). For production, specify data sources, retention windows, and privacy boundaries; place cameras with serviceable covers and clear indicator lights.
Fail‑Safes & Fallbacks
Design for AR/HUD outage: promote critical diegetic indicators (speed, attitude/heading, gear) and keep small physical instruments as last resort. If autonomy stack faults, revert to manual with an assertive cue and remove misleading overlays. If sensors degrade (heavy rain, dust), show degraded‑mode symbology and reduce reliance on world‑locked graphics. Keep manual sighting marks (simple reticles or pitch ladders) etched or printed as passive backups.
Sensor & Compute Placement
AR and autonomy ride on sensors: stereo or lidar at the header, radar in the nose, INS in the center, wheel encoders/airspeed at the chassis. Integrate them into A‑pillar fairings or header pods inside the wipe/defog zone. Provide washers/heaters, hydrophobic coatings, and vibration isolation. Compute nodes need thermal paths, EMI shielding, and redundant power buses with graceful brownout behavior.
Workflow: From Thumbnail to Callout Sheet
At thumbnail scale, draw the eyebox and sketch AR overlays as light lines aligned to the world. Show mode bands on structure. In orthos, add focal distance notes, latency targets, and declutter rules. In callouts, specify combiner area, waveguide apertures, sensor FOVs, mount stiffness, and service covers. Provide test scenes (sun glare, rain at night, dust, snow) and expected UI behavior in each.
Maritime, Ground, and Air Differences
Maritime bridges favor parallax‑free range ladders on window edges and berthing cues at bow corners; salt fog demands robust coatings. Ground vehicles emphasize lane/path overlays, gap acceptance, and low‑speed maneuvering cues; heavy rain and tunnels tax sensors—use confidence fades. Air/VTOL cockpits need attitude/velocity vectors, flight path markers, and landing cue discs; washout and reflections are acute—use anti‑reflective glass and sunload limits.
Accessibility and Cognitive Load
Design for color‑vision variance: encode meaning with shape and motion as well as hue. Provide read‑aloud and haptic echoes for key alerts. Keep glance legibility: critical cues must be readable within 100–200 ms. Use progressive disclosure to reveal details on demand without crowding the world view.
Maintenance & Modularity
Treat AR optics and diegetic light engines as LRUs with captive fasteners and keyed connectors. Provide wipe/defog service without removing structural rings. Include re‑calibration flows after windshield or camera replacement: printed fiducials or test patterns in the bay; on‑road calibration routes. Document cleaning agents safe for coatings.
Privacy, Security, and Ethics
Cameras and logs raise concerns. Provide recording indicators, data minimization, and consent affordances for passengers. Authenticate over‑the‑air updates; isolate safety‑critical networks; and design fail‑secure behavior (degrade gracefully, never spoof the world).
Rendering Language for Artists
Use translucent frusta to show HUD/AR volumes and ghost the structural ring beneath. Draw world‑locked lines with subtle parallax cues; screen‑locked widgets should hug the glass frame. Indicate mode bands with thin, continuous diegetic strips that are visible in peripheral vision. Add confidence halos and priority fades to teach readers how overlays behave. Include a night vs. day inset to demonstrate brightness strategy.
Concept‑to‑Production Handshake
End your packet with: eyebox dimensions; focal distances and apertures; latency budgets; color/brightness ranges; autonomy mode definitions and transitions; sensor list and FOVs; compute placement and power; failover/fallback logic; calibration procedures; IP/EMI specs; and maintainability notes. These turn narrative intent into a verifiable plan.
Case Studies in a Paragraph
A desert scout truck uses a beltline mode band and A‑pillar proximity ladders. AR paints a conformal path over dunes with confidence fading in blowing sand; autonomy announces “cresting left” with a yoke haptic and HUD banner. A VTOL cockpit shows a landing disc conformal to the pad with a clearance halo; diegetic rim LEDs echo envelope protection. A harbor bridge displays range ladders on window edges and projects a virtual tug line with tension readout; autonomy shifts to supervised near the pier, requiring the helmsman’s squeeze confirm for thruster commands.
Final Encouragement
Diegetic UI and AR make the cabin into the instrument, and autonomy turns intent into motion. If your pages declare eyes, layers, and authority—and show optics, sensors, and fail‑safes—your bridges and cockpits will feel confident under sun, rain, and stress. Draw the world as the screen, the structure as the status light, and the mode as a ring you can always see without looking—and both artists and engineers will believe it.